Claude

Before you can use Claude as an AI provider, set up an account, workspace, and API key on the Anthropic Console.

1. Create an account

• Go to the Anthropic Console.
• Sign up using an organization account rather than a personal account (recommended).
• Set up billing: API usage is billed separately from any Claude.ai subscription.

2. Create a workspace

Anthropic organizes usage and billing through workspaces.

• Create a new workspace within your organization.
• Use a clear naming convention, for example ai-studio-production or ai-studio-development.

Workspaces let you isolate environments (dev/test/prod), track usage per application or team, and apply spend limits per workspace.

3. Create an API key

• Generate an API key within the workspace.
• Copy the key immediately: it’s only shown once.

Secure storage

• Store API keys in a secure location, such as a secrets manager or encrypted storage.
• Never hardcode API keys in source code.
• Rotate keys periodically, especially for production environments.

Within Any2Info, API keys are stored separately from connection definitions to prevent accidental exposure: see Connections.

Note

Claude agents always authenticate through a connection of type API headers with exactly one header, holding the API key. Collection-based authentication is not supported.

4. Available models

Select the model when creating an agent.

Model	Best for
Claude Fable 5.1	The most demanding reasoning and long-horizon agentic work.
Claude Opus 5.5	Fable 5.1-level performance at lower cost, for long-running agentic coding and knowledge work.
Claude Opus 5	Demanding reasoning, coding, and agentic work.
Claude Sonnet 5	A balance between intelligence, speed, and cost. Suitable for most tasks and the default.
Claude Haiku 4.5	The fastest and most cost-effective model.

The Effort setting controls how much reasoning the model applies. Higher levels (up to Max) generally mean longer requests and more tokens used. It is not available for Claude Haiku 4.5.

5. Logging

Logging provides insight into how Claude is being used, and helps with debugging and compliance.

Typical logging includes request metadata, the model used, token usage, and errors and failures. It should not include full document contents (unless explicitly required) or personal/sensitive data without proper justification.

6. Usage and cost monitoring

The Anthropic Console provides usage and cost metrics per workspace. Monitor:

• Token usage per model
• Cost per workspace or environment
• Spikes in usage that may indicate misconfiguration

Set spend limits per workspace where possible, use lighter models such as Haiku for simple analysis tasks, and review usage periodically as agents are added or modified.

Next steps

Once your Anthropic account and workspace are set up, create a connection in Data Studio, select Claude as the provider, and continue to creating your first agent.